A basic requirement of a high availability distributed device system, such as a disk file subsystem, is to manage its data in an appropriate manner when the power supply is removed. In the case of short duration power interruptions, (less than 30 milliseconds) a subsystem would be expected to continue operating normally. If a longer power supply interruption occurred then the subsystem should shut down in a controlled manner without any loss of data. In the latter case, the system would be expected to continue running for a period of around 300 seconds while any data residing in temporary storage, such as cache memory, was written to disk. Normally the latter requirement has been achieved using UPS (uninterruptible power supply) technology to protect the complete disk file subsystem or the storage controller.
This approach has disadvantages. The number of UPS devices which can be economically and practically employed in a subsystem is limited by their cost and size. In addition, the UPS is dependant upon battery technology which may need to be maintained during the life of the product. Also, it is difficult to integrate the operation of the UPS within the disk file subsystem.
Therefore, it would thus be desirable to have a technology for the efficient delivery of standby power to manage the controlled shutdown of a device subsystem, such as a storage subsystem, when primary power has been lost.